Hi,
After inspection it looks like I am duplicating roman's and thorsten's
effort. However I am still not satisfied by their patch. There is still
an issue open with their patch which concerns the handling of errno
which is absolutely not thread safe in the context of green-threads (at
least). All syscalls must be protected in that case. I accept to take
this new target as something we can work on though...
Regards,
Guilhem.
Guilhem Lavaux wrote:
Hi,
Ok. Here is an update of the previous patch. After some discussion that
one should be compatible with AICAS' and kaffe's needs. I have slightly
abstractized the syscall interfaces for file's IO. Now I provide an
interface for doing jlong math and handling errors. For most
architecture the definition will only be these dummy macro. However in
some cases they may be useful (for compiler not supporting 64 bits for
example). A few function has been left in the original file but they
should be moved to the new interface too (mmap and lock).
This is the a preliminary patch as I am going now to "abstractize" the
network layer.
Regards,
Guilhem.
_______________________________________________
Classpath mailing list
Classpath@gnu.org
http://lists.gnu.org/mailman/listinfo/classpath